Automated colour segmentation of Tuberculosis bacteria thru region growing: A novel approach

Abstract

Medical image analysis is very challenging due to idiosyncrasies of medical profession. Object recognition with data mining techniques has helped doctors in case of medical emergencies for the image analysis, pattern identification and treatment. Over 180 million people died and more than one third of the population is carrier of Mycobacterium Tuberculosis (TB) bacteria as per the WHO statistics [1-5]. Segmentation of TB from the stained background is very challenging due to noise and debris in the image. In this paper, an automated segmentation of tuberculosis bacterium using image processing techniques is presented. Colour segmentation with region growing watershed algorithm is proposed for the bacterial identification.